I''ve already looked through the www.gameai.com website, but I''m just wondering really what games use fuzzy logic for some part of their AI. Is it fairly prevalent? What games (past, current, and upcoming) use fuzzy logic and how is it used? It is used to create uncertainty? Is it used in a FuSM? Thanks for any answers you guys can give.

I''m using it in my game... but I can''t tell you how just yet. (It''s a secret!)

Dave Mark - President and Lead Designer
Intrinsic Algorithm - "Reducing the world to mathematical equations!"
RIPPL Sports - NFL Statistical Analysis and Prediction System

quote: Original post by ares32585
I''ve already looked through the www.gameai.com website, but I''m just wondering really what games use fuzzy logic for some part of their AI. Is it fairly prevalent? What games (past, current, and upcoming) use fuzzy logic and how is it used? It is used to create uncertainty? Is it used in a FuSM?

I am really interested in fuzzy logic myself, but I don''t believe it''s very prevalent in games at all. This is confused by the fact that a lot of developers claim they are using a fuzzy state machine when in fact they are just using a non-deterministic state machine. This comes from people''s misunderstanding of ''fuzzy'', I guess. Maybe you misunderstand it too, as you mention uncertainty in your post - sorry if I''ve misjudged you though! Fuzzy logic isn''t uncertain or random, it is just a slightly different way of thinking. It is related to probability, but it''s not the same thing.

As one example of a game-related application, I expect fuzzy logic would be useful for creating some sort of steering behaviour or flocking algorithm. A fuzzy logic statement such as "if other unit is close, move away from other unit" would keep 2 units apart without there appearing to be some sort of hard limit on how close the 2 units can get. Add in fuzzy rules like "if other unit is distant, move closer to other unit" and "if leader unit is travelling in a different direction, turn towards that direction", and you probably have the basis of a very workable flocking or following system. Of course, you need to work out how to specify the fuzzy rules, but that''s not too difficult once you know how.

[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost | Asking Questions | Organising code files ]

What I meant by uncertainty is pretty much what you say here: "if other unit is close, move away from other unit" There is no "hard-coded" distance that the until will move away; there is no "certain" distance that it will move (in that sense it is uncertain). Does fuzzy logic indirectly introduce uncertainty into decisions and actions-I know that fuzzy logic itself isn't uncertain. If I'm still incorrect in my understanding, please let me know.



[edited by - ares32585 on July 31, 2002 11:27:54 PM]

umm, fuzzy logic is merely a system in which fractional values can be used to obtain a trueness or falseness. for instance to weigh certain events over other events. there MUST be a limit to what is defined as close (ie true). this can be used to adjust in smaller increments if the truth value is within tolerences. many games use fuzzy logic, however some give special names to how the fuzzy logic is used.

its true some game add randomness to the mix, but this dont mean they are not using fuzzy logic. just like a non deterministic state machine does not need to have randomness. randomness is ussually added so the ai unit is not perfect and exhibits flaws found in humans such as misfires and bad timing.

simply put:

quote:
An extension of two-valued logic such that statements need not be true or false, but may have a degree of truth between 0 and 1. Such a system can be extremely useful in designing control logic for real-world systems such as elevators.

its nothing more nor nothing less. there is a certain distance that is optimal. this is considered the most true distance (1.0), while farther you get distances would go away from 1.0 towards zero. unfrortunatly this is not useful since getting farther or closer means going towards 0. instead one would use 0.0 as the optimial distance "truth" value, and values away from it will be considered less optimal. thus an agent tries to keep as close to 0.0 as possible to maintain its optimal distance. you have to dictate this distance. you have to decide the threshholds.

a person : yeah that was how I understood fuzzy logic to work.

In a FuSM I believe that movement between states is probabilistic as compared to deterministic and non-deterministic state machines, however I''m not sure whether this is the same as ''fuzzy logic'' itself (although I''m sure fuzzy logic deals in probabilities as well).

I believe Swat 2 used fuzzy logic. It is fairly old, and I''ve never played it, but you might want to look it up. I think they used to to determine how the bad guys were going to act, based on stuff like how scared they were, how much danger they were in, etc

Trying is the first step towards failure.

If you''re looking for examples, www.gameai.com is a good ressource. Look into the current games section. The most notable examples I found were Civilization: Call To Power, and the Sims.

Cédric

What''s the difference between a non-deterministic state machine and a fuzzy state machine?

Is it that in a non-deterministic state machine, it can belong to more than one state at once, but in a fuzzy state machine, it can belong to more than one state at once to a certain degree?

I wonder myself what''s the differance beetwen a non deterministic fsm and a fuzzy logic machine.
I mean, in term of alghoritms the difference is clear, but can anybody really distinguish a game based on random fsm from the one based on fuzzy logic? if so, why?
I have read that fuzzy logic is particulary suitable for human charachter.
What about the SIM?